H3-World: Turning Video Generators into Interactive World Models

Danze Chen · hf · 2026-09-02

H3-World framework turns the MiniMax-H3 video generator into an interactive world model. It leverages the emerging language control capabilities in large video generators. By using structured instructions and temporal attention routing, it achieves precise character and camera control with minimal adaptation.

Original post →

More from Multimodal

Multimodal channel →